How artificial intelligence made a metal alloy easier to 3D print
Researchers at Washington State University used artificial intelligence to 3D print GRCop-42, a metal alloy developed by NASA, with laser power reduced by 80%. The result could make this technology accessible to smaller laboratories and companies as well.
An alloy that is difficult to print
GRCop-42 is a strategic alloy for aerospace applications, but its 3D printing requires expensive equipment and very powerful lasers.
GRCop-42 is an alloy composed of copper, chromium, and niobium. NASA developed it to produce components for combustion devices.
Its characteristics make it valuable: good fatigue resistance, high thermal conductivity, and excellent creep resistance. But 3D printing this alloy is extremely difficult.
Previous attempts required very powerful lasers, specialized equipment, and considerable resources. Long times and high costs have limited access to this technology.
- First successful configuration with only 500 watts of laser power
- Six printing configurations identified in three months
- 40 experiments conducted against the 37 previous failed attempts
- Reduction of the 80% in the power required compared to standards
The artificial intelligence approach
The researchers developed a predictive model that learns from failures to quickly identify working printing configurations.
The team from the School of Mechanical and Materials Engineering had accumulated 37 unsuccessful attempts. This data became the basis for training an artificial intelligence model.
The model was designed to predict which configurations could work. The negative results of the new tests also helped improve the predictions.
Jana Doppa, who led the study, explained the challenge: “It is a very difficult case for AI. With each attempt we essentially get a binary signal of success or failure.”.
The goal was to minimize the number of trials required. The system had to quickly reach the rare working configurations.
The results in three months
In just three months the team achieved more successes than all previous attempts, identifying six valid configurations with different power levels.
The team conducted 40 experiments in three months. The result: six configurations that allow printing GRCop-42 using different laser power levels.
The most significant milestone is printing with only 500 watts of power. It is the first time that GRCop-42 is printed with such low power.
The reduction in required laser power could make the GRCop-42 alloy accessible to universities, companies, and smaller laboratories, eliminating the need for specialized, bulky, and expensive equipment.
Future perspectives
The AI-based approach could be applied to other difficult-to-print alloys, reducing development time and costs.
The result significantly reduces the energy consumption required for printing. The ultimate goal is to eliminate the need for specialized equipment.
The developed methodology could be applied to other complex materials. Artificial intelligence accelerates the optimization process that would traditionally require years of trial and error.
The full press release from Washington State University is available for further technical details.
article written with the help of artificial intelligence systems
Q&A
How much laser power is now needed to print the GRCop-42 alloy?
Thanks to artificial intelligence, the GRCop-42 alloy can be printed with just 500 watts of laser power. This represents an 80% reduction compared to the power traditionally required.
Who developed the artificial intelligence model for this research?
The model was developed by researchers at Washington State University, led by Jana Doppa from the School of Mechanical and Materials Engineering. They used data from 37 previous failed attempts to train the system.
What are the main characteristics of the GRCop-42 alloy?
The GRCop-42 alloy, composed of copper, chromium, and niobium, offers high thermal conductivity, excellent creep resistance, and good fatigue resistance. It was developed by NASA specifically for aerospace combustion devices.
How many experiments were needed to find the successful configurations?
The team conducted 40 experiments over three months using the AI-based predictive model. This approach allowed them to identify six working printing configurations, surpassing the 37 failures accumulated previously.
Why was 3D printing of GRCop-42 previously difficult?
In the past, printing this alloy required very powerful lasers, specialized equipment, and considerable resources, limiting access to a few laboratories. High costs and technical complexity made the process inaccessible for smaller companies.
